Among older adults, osteoporosis and back pain are common alone, and also in combination. These conditions can cause enormous personal suffering and societal burden. A mainstay of physiotherapy treatment of back pain is the category of hands-on treatments known as 'manual therapy'. These are routinely used to assess and treat back pain in various clinical settings but their safety, efficacy, and mechanism of action has not previously been studied in individuals with back pain and osteoporosis.

The objectives of this thesis were to investigate, (1) the evidence for the effectiveness of manual therapy for spinal pain, (2) physiotherapists' perceptions and practice behaviors with respect to the use of manual therapy on individuals with osteoporosis, (2) the safety of posteroanterior (PA) spinal mobilization in the osteoporotic spine, (3) detection and determinants of spinal fracture under a PA load, and (4) whether PA stiffness can predict intervertebral range of motion (ROM) and flexibility in the cadaveric midthoracic spine of older adults.

To achieve these objectives I used a variety of research methods that included;​ a systematic review, a survey (171 physiotherapists), biomechanical testing, plain radiography, computed tomography (CT), dual energy X-ray absorptiometry, ash weight, and micro-computed tomography (12 cadaveric spine segments, one intact cadaver, 2 physiotherapists, 7 participants). The systematic review indicated that:​ (i) physiotherapy that included manual therapy at a dose of 30-45 minutes per session, for 4-8 weeks was effective in adult populations with back or neck pain, and (ii) clinically relevant differences between the manual therapy interventions used in clinical trials may influence the outcomes. My survey of physiotherapists found that 91% of respondents were concerned about fracture as a complication of treatment when using manual therapy in patients with osteoporosis.

Among the key findings in my study of vertebral biomechanics using cadaveric midthoracic spine segments was that vertebral body injury is an unlikely complication of PA mobilization in the midthoracic spine. Simulated PA mobilization using a mechanical testing machine produced spinous process fractures in every case and no vertebral body fractures. There was a reasonable margin between the failure load, in vitro, and the applied mobilization load, in vivo, for most specimens, however the lowest fracture thresholds (200N) approached the same force as the upper range of the applied loads (223N). Bone mineral density (BMD) of the whole vertebra was not a good predictor of PA failure load (r=0.18, p=0.68) but micro-CT measures of regional bone volume fraction of the spinous process base and middle regions, the sites of fracture, were strongly correlated with PA failure load (base:​ r=0.74, p=0.01;​ mid:​ r=0.73, p=0.01). Plain radiography and CT had poor sensitivity for these spinous process fractures (detected 3/12 and 6/12 fractures, respectively).

DXA scanning was an appropriate surrogate measure for thoracic spine segment bone mineral measurement. Volumetric BMD can be accurately estimated using the elliptical cylinder method. Trabecular thickness differs significantly between the spinous process and lamina regions, and may have influenced the site of fracture (p=0.003).

Cadaveric studies of spine kinematics during PA mobilization showed that the mobilized thoracic vertebra moves into extension as a result of the mobilization. Further, in cadaveric midthoracic spine segments from older adults, PA stiffness is inversely correlated with ROM and flexibility at the level at which the PA mobilization is applied (r= -0.78 - -0.90).